I think you have to talk about the pros and cons overall not just the amount.
Pros:
+ Can boost developers' morale. Feeling of contribution of their work to society is a plus for many people. If developers are shameful of their work or do not want it publicized, however, it could decrease morale.
+ Possible increase in amount of eyes on code and users of product, which may increase the need for code changes that could increase the quality of the product. The predicted quality increase is still often overestimated.
Cons:
- Most projects with any respectable amount of use in-the-wild (many are not used, and it is not a problem) require a level of support which might not have been necessary had they not been open-sourced. And the amount of time required for support and related added functionality is often underestimated.
- Despite best efforts, projects will inadvertently leak private company data. Usually the risk associated with data going public is small, but web searches can often find passwords, etc. Many things can mitigate this risk (thoughtful design and practices, a more secure network, etc.), but it is a risk.
Tips:
* The more time spent on fewer projects, the more successful they will be, and the more this will reflect positively on the company.
* You need a good balance where there are enough leads on the project to sustain some leaving, but too much complexity will kill a project.
* Company and organizational/cultural politics can severely weaken a project. This is probably just as valid of a reason that many large academic projects stink as the reason that they are not money-driven or having to deal with competition.
* Lawyers and legal/licensing concerns can kill or severely stagnate an OSS project.
Despite all that, I still think companies should do as much OSS work as possible, even if it isn't always good for business.